[ros-users] remote control (kbd / joystick / etc.) (brice rebsamen)

Ken Conley kwc at willowgarage.com
Fri May 27 17:20:23 UTC 2011


Hi Chad,

These are good points.  I encourage you to look as several of the new
top-level pages the community has added to the wiki in the past year
and suggest specific ways/areas they could be improved further.  These
pages include:

Library/functionalities:
http://www.ros.org/wiki/APIs

Sensors:
http://www.ros.org/wiki/Sensors

Robot-specific landing pages:
http://www.ros.org/wiki/Robots

Tools:
http://www.ros.org/wiki/Tools

It's really the robot-specific landing pages I'm most excited about
and think the community can contribute the most to.  I'm hoping that
the pages like:

http://www.ros.org/wiki/Robots/NXT

will be a place where NXT users can start answering more specific "how
do I do X with my robot" questions.

One of the reasons we are focused on TurtleBot is it is simply too
difficult to provide a generic "here are cool libraries" without
understanding what hardware the person is using, i.e. "navigation" has
a very different meaning if I have a Create, AscTec quadrotor, or
autonomous car.

cheers,
Ken


On Thu, May 26, 2011 at 12:41 PM, Jenkins, Odest Chadwicke
<odest_jenkins at brown.edu> wrote:
> Hi Ken,
>
> I think you make a very good point about developing a teleop_mobile
> stack.  That would certainly be a welcomed contribution.  I also agree
> that asking Willow to take this on is not the best use of time and
> effort.
>
> A larger point I am making is that we are starting to see more
> reinvention and refragmentation among the efforts of the ROS
> community.  I would attribute this circumstance to the lack of a
> global picture of ROS that people (new and established) in ROS can
> understand.  We often have to climb up the learning curve by scouring
> the more detail-oriented content of the wiki and various repositories.
>  This could be a dealbreaker for many types of people we would like to
> bring into the ROS community: app-level developers, people who design
> systems with usability and value in mind, etc.
>
> I think the answer to Brice's question (and similar questions) should
> be more obvious.  By an order of magnitude, ROS has been a great
> contribution to robotics, as an applications-layer protocol, message
> structure, and development environment.  However, ROS is still very
> far from enabling robots to provide value for real users and app
> developers.  More guidance from the ROS leadership as well as
> discussion with the current ROS community would help in broadening the
> ROS community in the future.
>
> (Brice, apologies for the threadjacking)
>
> -Chad
>
> On Thu, May 26, 2011 at 4:31 PM, Ken Conley <kwc at willowgarage.com> wrote:
>> If someone was willing to coordinate/maintain a "teleop_mobile" stack,
>> we would happily accept/anoint it.  As Brian notes, the difficulty is
>> in ensuring that such a 'general' teleoperation package generically
>> controls a variety of robots.  We would not be able to do such a stack
>> ourselves (at least for Electric) as our post-ICRA todo list is a bit
>> too much right now.
>>
>> It sounds like there are (at least) two good starting points for
>> packages to include.  For keyboard, the stack that Chad mentions:
>>
>> http://www.ros.org/wiki/teleop_twist_keyboard
>>
>> And for joystick, we have our teleoperation package we use with the TurtleBot:
>>
>> http://www.ros.org/wiki/turtlebot_teleop
>>
>> The joystick case is a bit more difficult as you also have to
>> parameterize a bit on the joystick.
>>
>> Our expectation for a maintainer would be to coordinate the community
>> to get good documentation in place, and also coordinate with the
>> community to test across multiple robot bases.
>>
>>  - Ken
>>
>> On Thu, May 26, 2011 at 9:17 AM, Jenkins, Odest Chadwicke
>> <odest_jenkins at brown.edu> wrote:
>>> Hi Brice,
>>>
>>> I believe you are correct in that base movement control has been
>>> reinvented several times over in ROS.  We wrote our own a while back,
>>> but there are probably other quality movement controllers across the
>>> ROS space:
>>>
>>>  http://www.ros.org/wiki/teleop_twist_keyboard
>>>
>>> teleop_twist_keyboard was based on the old playerjoy utility from
>>> Player, which includes a stop command and has limited handling of key
>>> press/release events.  We often use teleop_twist_keyboard for the
>>> Create, AR.Drone, and PR2 (as in the following video):
>>>
>>>  http://www.youtube.com/watch?v=M-9sDNnGtIs
>>>
>>> I think your post is great reminder that the ROS community could
>>> benefit from a clearer organization of packages, messages, and
>>> functionality in ROS.  Such a clear organization does not seem likely
>>> to happen organically without some guidance from the ROS leadership.
>>>
>>> -Chad
>>> _______________________________________________
>>> ros-users mailing list
>>> ros-users at code.ros.org
>>> https://code.ros.org/mailman/listinfo/ros-users
>>>
>>
>



More information about the ros-users mailing list